Author name cluster

Jiaming Tian

Possible papers associated with this exact author name in Arrow. This page groups case-insensitive exact name matches and is not a full identity disambiguation profile.

2 papers

1 author row

AAMAS Conference 2026 Conference Paper

GLEAR: A Graph Logic-Enhanced RAG Framework for Legal QA

Jingyun Sun
Jiaming Tian
Jie Shi
Yixin Zhang
Wenxi Sheng
Yang Li

Existing Legal Large Language Models (3LMs) can answer user queries due to their parameterized ability to understand and generatelegaltext. However, theylackeffectiveutilizationoflogicallegal knowledge, limitingtheirperformanceinlegalQuestionAnswering (QA). To leverage the semantic understanding capability of 3LMs while effectively capturing the logical relationships between legal knowledge, we propose a graph logic-enhanced RAG framework for legal QA, named GLEAR. The framework first structures various legal knowledge into a multi-source heterogeneous knowledge graph, and then enhances the model’s response quality through three core processes: dual-driven legal knowledge retrieval, key logical path mining, and inference enhancement. Experimental results show that GLEAR outperforms the baselines by an average of 14 percentage points across five traditional legal NLP tasks. In the free-form legal QA task, GLEAR also significantly surpasses the baselines in terms of response accuracy, professionalism, and comprehensiveness. Additionally, experiments demonstrate that GLEARoutperformsthestandardRAGmethodinbothperformance and computational efficiency.

PDF

NeurIPS Conference 2025 Conference Paper

Table as a Modality for Large Language Models

Liyao Li
Chao Ye
Wentao Ye
Yifei Sun
Zhe Jiang
Haobo Wang
Jiaming Tian
Yiming Zhang

To migrate the remarkable successes of Large Language Models (LLMs), the community has made numerous efforts to generalize them to the table reasoning tasks for the widely deployed tabular data. Despite that, in this work, by showing a probing experiment on our proposed StructQA benchmark, we postulate that even the most advanced LLMs (such as GPTs) may still fall short of coping with tabular data. More specifically, the current scheme often simply relies on serializing the tabular data, together with the meta information, then inputting them through the LLMs. We argue that the loss of structural information is the root of this shortcoming. In this work, we further propose TAMO, which bears an ideology to treat the tables as an independent modality integrated with the text tokens. The resulting model in TAMO is a multimodal framework consisting of a hypergraph neural network as the global table encoder seamlessly integrated with the mainstream LLM. Empirical results on various benchmarking datasets, including HiTab, WikiTQ, WikiSQL, FeTaQA, and StructQA, have demonstrated significant improvements on generalization with an average relative gain of 42. 65%.

PDF Details